INTRODUCTION
Food insecurity, defined as uncertain or limited access to nutritionally adequate and safe foods, can negatively impact health outcomes and healthcare utilization of those with multiple chronic conditions (MCC) 1–3. However, few studies have examined food insecurity among older adults with MCC in primary care. This study examines socio-demographic and clinical factors associated with food insecurity among older adults with MCC in primary care in order to direct food insecurity screening efforts to high-risk populations.
METHODS
From March to December 2016, we conducted a cross-sectional telephone survey of adults aged 60 years or older receiving care from an urban academic primary care practice. Inclusion criteria included fluency in English, Spanish, or Chinese (Cantonese/Mandarin); at least two concurrent chronic conditions based on diagnosis codes from the Elixhauser Comorbidity Index 4; and at least one clinic visit in the last year. We identified eligible participants from the electronic health record (EHR), selected participants using random clinic sampling stratified by race/ethnicity, and recruited participants using an opt-out letter. The survey included the 10-item U.S. Adult Food Security Survey Module (food insecurity defined as 3+ affirmative responses), education, employment, annual household income, English proficiency, cost-related medication non-adherence, self-reported health status, and utilization of community food resources and the Supplemental Nutrition Assistance Program (SNAP). We linked survey responses to the following variables from the EHR: age, sex, race/ethnicity, insurance type, preferred language, ICD-10 diagnosis codes, and healthcare utilization. We conducted bivariate analyses and a parsimonious multivariate logistic regression model, adjusting for socio-demographic and clinical factors that were significantly different between older adults with and without food insecurity. We did not include the number of medications since it was moderately correlated with the number of chronic conditions (r = 0.33).
RESULTS
The prevalence of food insecurity in this sample of older adults with MCC (n = 475) was 8.2%. The food insecure participants were younger (mean 68 versus 73 years old) and more likely to identify as female, disabled, and African American (Table 1). They also had lower levels of education and income. Among food insecure participants, 12% reported SNAP enrollment and 69% reported utilization of at least 1 community food resource. Food insecure participants had higher burden of chronic disease (46% versus 21% had 5+ chronic conditions) compared with food secure participants. In addition, food insecure participants were more likely to report at least 1 behavioral health diagnosis (defined as alcohol or drug abuse, psychoses, or depression, 67% versus 26%), polypharmacy (17 versus 12 total medications), and more primary care visits in the last year (5.3 versus 3.5 visits) compared with food secure participants. In a multivariate model (Table 2), factors positively associated with food insecurity were being African American compared with being White (AOR 5.8 95%CI (2.2, 15.6)) and having at least one behavioral health diagnosis (AOR 4.1 95%CI (1.9, 9.0)). Older age was negatively associated with food insecurity (AOR 0.92 95%CI (0.87, 0.98)).
Table 1.
Socio-demographic and Clinical Characteristics by Food Insecurity Status of 475 Older Adult Patients with Multiple Chronic Conditions from an Urban Academic Primary Care Practice
| Total study population n = 475 |
No food insecurity n = 436 | Food insecurity n = 39 |
p value* | |
|---|---|---|---|---|
| Socio-demographic characteristics | ||||
| Age, years† | ||||
| Mean ± SD, median (range) | 72.2 ± 7.8, 71 (58–98) | 72.6 ± 7.8, 72 (60–98) | 67.7 ± 6.2, 66 (58–87) | < 0.001 |
| 51–60 | 1.5% | 1.4% | 2.6% | |
| 61–70 | 45.5% | 42.9% | 74.4% | |
| 71–80 | 36.0% | 37.6% | 17.9% | 0.0014 |
| > 81 | 17.1% | 18.1% | 5.1% | |
| Male† | 49.1% | 50.5% | 33.3% | 0.04 |
| Race/ethnicity† | ||||
| Non-Hispanic White | 54.3% | 55.5% | 41.0% | |
| Asian | 22.5% | 23.6% | 10.3% | < 0.001 |
| African American | 7.6% | 5.3% | 33.3% | |
| Latino | 9.1% | 8.7% | 12.8% | |
| American Indian | 0.8% | 0.7% | 2.6% | |
| Other/unknown | 5.7% | 6.2% | - | |
| Preferred language† | ||||
| English | 84.8% | 84.6% | 87.2% | |
| Chinese** | 9.9% | 10.3% | 5.1% | 0.59 |
| Spanish | 4.8% | 4.6% | 7.7% | |
| Other | 0.4% | 0.5% | 0.0% | |
| Limited English proficient | 17.3% | 17.2% | 17.9% | 0.91 |
| Education | ||||
| < HS diploma | 14.3% | 14.2% | 15.4% | |
| HS diploma to some college | 24.8% | 22.2% | 53.8% | < 0.001 |
| ≥ College degree | 60.2% | 63.1% | 28.3% | |
| Missing | 0.6% | 0.5% | 2.6% | |
| Employment | ||||
| Full or part time | 22.3% | 23.2% | 12.8% | |
| Disabled | 5.7% | 3.7% | 28.2% | < 0.001 |
| Retired/unemployed/other | 72.0% | 73.2% | 59.0% | |
| Total household size | ||||
| Mean ± SD, median (range) | 2.2 ± 1.2, 2 (1–9) | 2.2 ± 1.2, 2 (1–9) | 1.8 ± 1.0, 1 (1–5) | 0.015 |
| Income | ||||
| ≤ $20,000 | 24.8% | 21.1% | 66.7% | |
| > $20,000 | 63.6% | 67.0% | 25.6% | < 0.001 |
| Missing/don’t know/not sure | 11.6% | 11.9% | 7.7% | |
| Any form of Medicare***† | 72.0% | 71.3% | 79.5% | 0.28 |
| Cost-related medication non-adherence | 13.7% | 10.3% | 51.3% | < 0.001 |
| Fair or poor health status | 31.5% | 28.9% | 61.6% | < 0.001 |
| SNAP use | 4.4% | 3.7% | 12.8% | 0.03 |
| Use of any community food resource^ | 17.5% | 12.8% | 69.2% | < 0.001 |
| Clinical characteristics | ||||
| Chronic condition diagnoses^^† | ||||
| Mean ± SD, median (range) | 3.6 ± 1.6, 3 (1–12) | 3.5 ± 1.5, 3 (1–12) | 4.5 ± 2.0, 4 (2–9) | 0.0011 |
| 0–1 chronic conditions | 1.7% | 1.8% | - | |
| 2–4 chronic conditions | 74.9% | 76.8% | 53.8% | 0.002 |
| 5+ chronic conditions | 23.4% | 21.3% | 46.2% | |
| Physical health diagnoses | ||||
| Congestive heart failure | 5.5% | 4.8% | 12.8% | 0.04 |
| Pulmonary circulation disease | 2.5% | 2.1% | 7.7% | 0.03 |
| Paralysis | 0.4% | 0.2% | 2.6% | 0.03 |
| Chronic pulmonary disease | 19.8% | 18.6% | 33.3% | 0.03 |
| Rheumatoid arthritis | 7.6% | 6.7% | 17.9% | 0.01 |
| Obesity | 21.9% | 20.6% | 35.9% | 0.03 |
| Behavioral health diagnoses | ||||
| Alcohol abuse | 2.7% | 2.3% | 7.7% | 0.05 |
| Drug abuse | 3.4% | 2.5% | 12.8% | < 0.001 |
| Psychoses | 4.0% | 3.0% | 15.4% | < 0.001 |
| Depression | 23.4% | 21.3% | 46.2% | < 0.001 |
| Any behavioral health diagnosis | 29.3% | 25.9% | 66.7% | < 0.001 |
| Number of medications† | ||||
| Mean ± SD, median (range) | 12.8 ± 6.6, 12 (1–37) | 12.4 ± 6.3, 12 (1–35) | 16.7 ± 9.0, 14 (1–37) | 0.003 |
| 0–1 | 0.4% | 0.2% | 2.6% | |
| 2–4 | 5.9% | 6.2% | 2.6% | |
| 5–7 | 17.3% | 17.7% | 12.8% | 0.11 |
| 8+ | 76.4% | 75.9% | 82.1% | |
| Body mass index, kg/m2† | ||||
| Mean ± SD | 27.8 ± 6.3 | 27.4 ± 6.0 | 31.7 ± 8.2 | < 0.001 |
| Median (range) | 26.4 (16.7–54.6) | 26.1 (17.1–54.2) | 30.8 (16.7–54.6) | |
| Underweight, < 18.5 kg/m2 | 1.5% | 1.4% | 2.6% | |
| Normal, 18.5–24.9 kg/m2 | 36.4% | 38.1% | 17.9% | 0.003 |
| Overweight, 25–29.9 kg/m2 | 33.1% | 33.7% | 25.6% | |
| Obese, ≥ 30 kg/m2 | 29.3% | 27.1% | 52.8% | |
| Healthcare utilization | ||||
| Primary care visits in last year† | ||||
| Mean ± SD, median (range) | 3.6 ± 3.0, 3 (0–20) | 3.5 ± 2.9, 3 (0–20) | 5.3 ± 3.6, 5 (0–17) | < 0.001 |
| ED visits in last year† | ||||
| Mean ± SD, median (range) | 0.28 ± 1.0, 0 (0–10) | 0.23 ± 0.77, 0 (0–8) | 0.85 ± 2.2, 0 (0–10) | 0.14 |
| Hospitalizations in last year† | ||||
| Mean ± SD, median (range) | 0.12 ± 0.32, 0 (0–1) | 0.12 ± 0.32, 0 (0–1) | 0.13 ± 0.34, 0 (0–1) | 0.80 |
*p value compares respondents with no food insecurity with respondents with food insecurity
**Chinese = Cantonese and Mandarin languages
***Medicare or Medicare advantage
EHR, electronic health record; HS, high school; SNAP, Supplemental Nutrition Assistance Program; ED, emergency department
^Community food resources include food pantry or food bank, soup kitchen, congregated meal site, church or community-based organization food distribution, Project Open Hand, and Meals on Wheels
^^This is an abbreviated list of chronic disease diagnoses from the Elixhauser Comorbidity Index. All other chronic disease diagnoses from the index (n = 20) were not significantly different between patients with and without food insecurity
†Data derived from the EHR
Table 2.
Adjusted Odds Ratios for Food Insecurity by Socio-demographic and Clinical Characteristics (the Model Was Adjusted for Age, Sex, Race/Ethnicity, Presence of Behavioral Health Diagnosis, Multiple Chronic Conditions, Body Mass Index, and Number of Primary Care Visits in the Last Year That Were Significantly Different between Older Adults With and Without Food Insecurity) Among 475 Older Adult Patients with Multiple Chronic Conditions from an Urban Academic Primary Care Practice
| Adjusted odds ratio for food insecurity (95%CI) | |
|---|---|
| Age | 0.92 (0.87, 0.98) |
| Female (ref: male) | 1.20 (0.55, 2.6) |
| Race/ethnicity (ref: White) | |
| African American | 5.8 (2.2, 15.6) |
| All others (Latino, Asian American, American Indian, other/unknown) | 1.24 (0.52, 3.0) |
| 1+ behavioral health diagnosis (ref: none) | 4.1 (1.9, 9.0) |
| 5+ chronic conditions (ref: 0–4) | 1.84 (0.78, 4.3) |
| Body mass index ≥ 30 kg/m2 (ref: BMI < 29.99 kg/m2) | 1.54 (0.70, 3.4) |
| Number of primary care visits in the last year | 1.06 (0.96, 1.18) |
DISCUSSION
Our study is among the first to examine how food insecurity is associated with clinical and healthcare utilization data of older adult primary care patients with MCC. In this population of older adults with MCC, the strongest risk factors for food insecurity were being African American, having a behavioral health diagnosis, and younger age.
We found a positive association between food insecurity with at least one behavioral health diagnosis. This finding has important clinical and population health implications. Among patients with diabetes, food insecurity has been associated with depression and stress, which can influence chronic disease management behaviors 5, 6. This study adds to our understanding of the complex relationships between food insecurity and behavioral health, particularly among older adults with MCC. Among these patients, focusing food insecurity screening efforts to those with a behavioral health diagnosis may efficiently identify patients who may benefit from a referral to food resources.
Study limitations include limited power and generalizability from one study site. Strengths include a multiethnic and linguistically diverse sample and a focus on MCC. There are emerging efforts to embed screening for health-related social needs, including food insecurity, into clinical care. Strategies are needed to identify at-risk populations to direct interventions. Our findings indicate that EHR data may allow for such a population health approach.
Acknowledgments
We thank Ying Wang, Michael Sharp, Marynieves Diaz-Mendez, and Filmer Yu for their contributions to data collection and management.
Funding Information
This research was supported by the National Institute on Aging (award numbers R03 AG050880 and P30 AG044281). Dr. Jih is supported by the National Center for Advancing Translational Sciences of the National Institutes of Health (award number KL2 TR001870).
Compliance with Ethical Standards
Conflict of Interest
Drs. Jih, Jin, Boscardin, and Ritchie received grants from the National Institutes of Health, during the conduct of the study. Dr. Seligman receives funding from Feeding America, a 501c3 non-profit dedicated to ending hunger in the USA, to serve as its Senior Medical Advisor.
Footnotes
This work was presented as a poster presentation at the Society of General Internal Medicine Annual Meeting on April 12, 2018 and at the American Geriatrics Society Annual Meeting on May 4, 2018.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Gundersen C, Ziliak JP. Food Insecurity And Health Outcomes. Health affairs (Project Hope). 2015;34(11):1830–1839. doi: 10.1377/hlthaff.2015.0645. [DOI] [PubMed] [Google Scholar]
- 2.Jih J, Stijacic-Cenzer I, Seligman HK, Boscardin WJ, Nguyen TT, Ritchie CS. Chronic disease burden predicts food insecurity among older adults. Public health nutrition. 2018:1-6. [DOI] [PMC free article] [PubMed]
- 3.Bhargava V, Lee JS. Food Insecurity and Health Care Utilization Among Older Adults. Journal of Applied Gerontology. 2017;36(12):1415–1432. doi: 10.1177/0733464815625835. [DOI] [PubMed] [Google Scholar]
- 4.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Medical care. 1998;36(1):8–27. doi: 10.1097/00005650-199801000-00004. [DOI] [PubMed] [Google Scholar]
- 5.Walker RJ, Campbell JA, Egede LE. Differential Impact of Food Insecurity, Distress, and Stress on Self-care Behaviors and Glycemic Control Using Path Analysis. Journal of general internal medicine. 2019. [DOI] [PMC free article] [PubMed]
- 6.Montgomery J, Lu J, Ratliff S, Mezuk B. Food Insecurity and Depression Among Adults With Diabetes: Results From the National Health and Nutrition Examination Survey (NHANES) The Diabetes educator. 2017;43(3):260–271. doi: 10.1177/0145721717699890. [DOI] [PMC free article] [PubMed] [Google Scholar]